Azure’s Machine Learning Offerings

Azure offers several services for working with machine learning (ML), and these span from general purpose to task specific. General-purpose services can be used to build the widest range of ML-based solutions, but they typically require longer initial development time as developers collect and format data, choose and configure algorithms, and tune experiments to build models for a solution. Task-specific services have a shorter ramp-up time. For example, a speech application will be faster to develop using a service that provides speech-based models compared with using a general-purpose service.

Azure ML offerings include the following:

Azure Machine Learning is the most general purpose of Azure’s ML offerings in that it provides the widest range of capabilities for building, training, and deploying models using first-party and third-party tools, frameworks, and languages.

Azure Cognitive Services is a collection of ML services that allow developers to add AI to applications with reduced upfront cost and expertise. Services address applications including vision, speech, natural language, anomaly detection, and search. Azure Cognitive Services solutions are less expensive to deploy and scale for these application types than comparable ML environments because the models are prebuilt and pretrained. Developers might choose Cognitive Services when they have some ML familiarity and need components for which most of the upfront building and training work is complete, but customizations are possible.